Hi there,

While I was playing with PostScript a few weeks ago, I found out
that it is possible to list the names of the keywords in PostScript's
dictionaries.

Because we're going to swich from an Apple LaserWriter to an
Agfa P3400PS Laser Printer, I thought it might be nice to have a way
to compare the internal dictionaries of these printers.

This resulted in a recursive PostScript program, systemdict.ps, that 
lists the total contents of the printer's dicts.

The output is as follows (in regular expression syntax):
  level keyword type value
  where:
    level 	= (.   )*
    keyword 	= <The human readable keyword>
    type	= "(" ((typeId ( access | none )) | none) ")"
    typeId	= (array|boolean|dict|file|fontId|integer|mark|name|
		   null|operator|packedarray|real|save|string)
    access	= (","(r|-)(w|-)(x|-))
    value	= (( "=" <the keyword's value>) | none )
    none	= ""		     

  which means:
    level gives the level of the dictionary:
      "" 	   means the upper level
      ".   "	   means one level below the upper level
      ".   .   "   means two levels below the upper level
    and so on

    Each time a dictionary is encountered the program descends a level
    and that dictionary is then listed.
    An empty line indicates that the program is ascending one level.
    For instance, a part of the systemdict might look like:
	type (operator)			% upper level (systemdict)
	userdict (dict,rw-)		% userdict is encountered
	.   #copies (integer) = 1	% #copies is a keyword in userdict
	.   $IdleTimeDict (dict,rw-)	% a new dict is encountered
	.   .   ROMnames (array,r--)	% an entry in $IdleTimeDict
					% empty line: ascending one level
	.   LocalDict (dict,rw-)	% LocalDict is a keyword in userdict..
	.   .   Dlen (integer) = 24	% .. which is again a dict
					% empty line: ascending
	.   start (packedarray,r-x)	% a keyword in userdict
					% empty line: back to systemdict
	usertime (operator)		% a keyword in systemdict
	version (string,r--) = 38.0	%   ...


    In the latter example the keywords are: 
      type, userdict, #copies, $IdleTimeDict, ROMnames, LocalDict, Dlen, 
      start, usertime and version.

    The types are:
      operator, dict, integer, array, packedarray and string.

    The accesses are:
      r = read
      w = write
      x = execute
    so rw- means: read access, write access, not executable 

    If the keyword's value is of the type integer, string, name, boolean
    or real, then the value of the keyword is listed if the ascii 
    representation doesn't exeed 19 characters. If it exeeds 19 characters 
    then the representation is clipped, which is indicated by three dots.
    For instance:
	Notice (string,r--) = Helvetica is a ...

    
The font dictionaries, located in the FontDictionary, contain a dictionary
called CharStrings. See the Red Book page 92 for a description of this
dictionary. The contents of these dictionaries is not listed (it would
produce many many pages of identical information). Instead, three dots 
indicate that there is more...
For instance:
	FontDirectory (dict,r--)
	.   Courier (dict,r--)
	.   .   CharStrings (dict,r--)
	.   .   .   ...

	.   .   Encoding (array,r--)
	.   .   FID (fontId)
	and so on


Every dictionary is first sorted in alphabetically ascending order by the 
procedure qsort, written August 31, 1989  by Scott Hemphill, who posted 
his routine to comp.lang.postscript a while ago as an addition to the 
PostScript toolkit.


Line 265 in the file systemdict.ps is:
/DictName (systemdict) def

The string systemdict can be changed to anything else that results in a 
dictionary on the stack. 
For instance:
	/DictName (userdict) def
or:
	/DictName (systemdict /statusdict get) def

If you do so, be sure to modify line 74 (starting level information) as well.


If you have any comments or if you encounter any bugs, please let me know.

Jos de Bruijne
Philips Research Laboratories
Eindhoven, The Netherlands
Email: debruyne@prles3.prl.philips.nl